/* $OpenBSD: eval.c,v 1.34 2009/01/29 23:27:26 jaredy Exp $ */ /*- * Copyright (c) 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010 * Thorsten Glaser * * Provided that these terms and disclaimer and all copyright notices * are retained or reproduced in an accompanying document, permission * is granted to deal in this work without restriction, including un- * limited rights to use, publicly perform, distribute, sell, modify, * merge, give away, or sublicence. * * This work is provided "AS IS" and WITHOUT WARRANTY of any kind, to * the utmost extent permitted by applicable law, neither express nor * implied; without malicious intent or gross negligence. In no event * may a licensor, author or contributor be held liable for indirect, * direct, other damage, loss, or other issues arising in any way out * of dealing in the work, even if advised of the possibility of such * damage or existence of a defect, except proven that it results out * of said person's immediate fault when using the work as intended. */ #include "sh.h" __RCSID("$MirOS: src/bin/mksh/eval.c,v 1.83 2010/02/25 20:18:15 tg Exp $"); /* * string expansion * * first pass: quoting, IFS separation, ~, ${}, $() and $(()) substitution. * second pass: alternation ({,}), filename expansion (*?[]). */ /* expansion generator state */ typedef struct Expand { /* int type; */ /* see expand() */ const char *str; /* string */ union { const char **strv; /* string[] */ struct shf *shf; /* file */ } u; /* source */ struct tbl *var; /* variable in ${var..} */ short split; /* split "$@" / call waitlast $() */ } Expand; #define XBASE 0 /* scanning original */ #define XSUB 1 /* expanding ${} string */ #define XARGSEP 2 /* ifs0 between "$*" */ #define XARG 3 /* expanding $*, $@ */ #define XCOM 4 /* expanding $() */ #define XNULLSUB 5 /* "$@" when $# is 0 (don't generate word) */ #define XSUBMID 6 /* middle of expanding ${} */ /* States used for field splitting */ #define IFS_WORD 0 /* word has chars (or quotes) */ #define IFS_WS 1 /* have seen IFS white-space */ #define IFS_NWS 2 /* have seen IFS non-white-space */ static int varsub(Expand *, const char *, const char *, int *, int *); static int comsub(Expand *, const char *); static char *trimsub(char *, char *, int); static void glob(char *, XPtrV *, int); static void globit(XString *, char **, char *, XPtrV *, int); static const char *maybe_expand_tilde(const char *, XString *, char **, int); static char *tilde(char *); #ifndef MKSH_NOPWNAM static char *homedir(char *); #endif static void alt_expand(XPtrV *, char *, char *, char *, int); static size_t utflen(const char *); static void utfincptr(const char *, mksh_ari_t *); /* UTFMODE functions */ static size_t utflen(const char *s) { size_t n; if (UTFMODE) { n = 0; while (*s) { s += utf_ptradj(s); ++n; } } else n = strlen(s); return (n); } static void utfincptr(const char *s, mksh_ari_t *lp) { const char *cp = s; while ((*lp)--) cp += utf_ptradj(cp); *lp = cp - s; } /* compile and expand word */ char * substitute(const char *cp, int f) { struct source *s, *sold; sold = source; s = pushs(SWSTR, ATEMP); s->start = s->str = cp; source = s; if (yylex(ONEWORD) != LWORD) internal_errorf("substitute"); source = sold; afree(s, ATEMP); return (evalstr(yylval.cp, f)); } /* * expand arg-list */ char ** eval(const char **ap, int f) { XPtrV w; if (*ap == NULL) { union mksh_ccphack vap; vap.ro = ap; return (vap.rw); } XPinit(w, 32); XPput(w, NULL); /* space for shell name */ while (*ap != NULL) expand(*ap++, &w, f); XPput(w, NULL); return ((char **)XPclose(w) + 1); } /* * expand string */ char * evalstr(const char *cp, int f) { XPtrV w; char *dp = null; XPinit(w, 1); expand(cp, &w, f); if (XPsize(w)) dp = *XPptrv(w); XPfree(w); return (dp); } /* * expand string - return only one component * used from iosetup to expand redirection files */ char * evalonestr(const char *cp, int f) { XPtrV w; char *rv; XPinit(w, 1); expand(cp, &w, f); switch (XPsize(w)) { case 0: rv = null; break; case 1: rv = (char *) *XPptrv(w); break; default: rv = evalstr(cp, f&~DOGLOB); break; } XPfree(w); return (rv); } /* for nested substitution: ${var:=$var2} */ typedef struct SubType { struct tbl *var; /* variable for ${var..} */ struct SubType *prev; /* old type */ struct SubType *next; /* poped type (to avoid re-allocating) */ short stype; /* [=+-?%#] action after expanded word */ short base; /* begin position of expanded word */ short f; /* saved value of f (DOPAT, etc) */ short quote; /* saved value of quote (for ${..[%#]..}) */ } SubType; void expand(const char *cp, /* input word */ XPtrV *wp, /* output words */ int f) /* DO* flags */ { int c = 0; int type; /* expansion type */ int quote = 0; /* quoted */ XString ds; /* destination string */ char *dp; /* destination */ const char *sp; /* source */ int fdo, word; /* second pass flags; have word */ int doblank; /* field splitting of parameter/command subst */ Expand x = { /* expansion variables */ NULL, { NULL }, NULL, 0 }; SubType st_head, *st; int newlines = 0; /* For trailing newlines in COMSUB */ int saw_eq, tilde_ok; int make_magic; size_t len; if (cp == NULL) internal_errorf("expand(NULL)"); /* for alias, readonly, set, typeset commands */ if ((f & DOVACHECK) && is_wdvarassign(cp)) { f &= ~(DOVACHECK|DOBLANK|DOGLOB|DOTILDE); f |= DOASNTILDE; } if (Flag(FNOGLOB)) f &= ~DOGLOB; if (Flag(FMARKDIRS)) f |= DOMARKDIRS; if (Flag(FBRACEEXPAND) && (f & DOGLOB)) f |= DOBRACE_; Xinit(ds, dp, 128, ATEMP); /* init dest. string */ type = XBASE; sp = cp; fdo = 0; saw_eq = 0; tilde_ok = (f & (DOTILDE|DOASNTILDE)) ? 1 : 0; /* must be 1/0 */ doblank = 0; make_magic = 0; word = (f&DOBLANK) ? IFS_WS : IFS_WORD; /* clang doesn't know OSUBST comes before CSUBST */ memset(&st_head, 0, sizeof(st_head)); st = &st_head; while (1) { Xcheck(ds, dp); switch (type) { case XBASE: /* original prefixed string */ c = *sp++; switch (c) { case EOS: c = 0; break; case CHAR: c = *sp++; break; case QCHAR: quote |= 2; /* temporary quote */ c = *sp++; break; case OQUOTE: word = IFS_WORD; tilde_ok = 0; quote = 1; continue; case CQUOTE: quote = st->quote; /* XXX correct? */ continue; case COMSUB: tilde_ok = 0; if (f & DONTRUNCOMMAND) { word = IFS_WORD; *dp++ = '$'; *dp++ = '('; while (*sp != '\0') { Xcheck(ds, dp); *dp++ = *sp++; } *dp++ = ')'; } else { type = comsub(&x, sp); if (type == XCOM && (f&DOBLANK)) doblank++; sp = strnul(sp) + 1; newlines = 0; } continue; case EXPRSUB: word = IFS_WORD; tilde_ok = 0; if (f & DONTRUNCOMMAND) { *dp++ = '$'; *dp++ = '('; *dp++ = '('; while (*sp != '\0') { Xcheck(ds, dp); *dp++ = *sp++; } *dp++ = ')'; *dp++ = ')'; } else { struct tbl v; char *p; v.flag = DEFINED|ISSET|INTEGER; v.type = 10; /* not default */ v.name[0] = '\0'; v_evaluate(&v, substitute(sp, 0), KSH_UNWIND_ERROR, true); sp = strnul(sp) + 1; for (p = str_val(&v); *p; ) { Xcheck(ds, dp); *dp++ = *p++; } } continue; case OSUBST: { /* ${{#}var{:}[=+-?#%]word} */ /* format is: * OSUBST [{x] plain-variable-part \0 * compiled-word-part CSUBST [}x] * This is where all syntax checking gets done... */ const char *varname = ++sp; /* skip the { or x (}) */ int stype; int slen = 0; sp = cstrchr(sp, '\0') + 1; /* skip variable */ type = varsub(&x, varname, sp, &stype, &slen); if (type < 0) { char *beg, *end, *str; unwind_substsyn: sp = varname - 2; /* restore sp */ end = (beg = wdcopy(sp, ATEMP)) + (wdscan(sp, CSUBST) - sp); /* ({) the } or x is already skipped */ if (end < wdscan(beg, EOS)) *end = EOS; str = snptreef(NULL, 64, "%S", beg); afree(beg, ATEMP); errorf("%s: bad substitution", str); } if (f & DOBLANK) doblank++; tilde_ok = 0; if (type == XBASE) { /* expand? */ if (!st->next) { SubType *newst; newst = alloc(sizeof(SubType), ATEMP); newst->next = NULL; newst->prev = st; st->next = newst; } st = st->next; st->stype = stype; st->base = Xsavepos(ds, dp); st->f = f; st->var = x.var; st->quote = quote; /* skip qualifier(s) */ if (stype) sp += slen; switch (stype & 0x7f) { case '0': { char *beg, *mid, *end, *stg; mksh_ari_t from = 0, num = -1, flen, finc = 0; beg = wdcopy(sp, ATEMP); mid = beg + (wdscan(sp, ADELIM) - sp); stg = beg + (wdscan(sp, CSUBST) - sp); if (mid >= stg) goto unwind_substsyn; mid[-2] = EOS; if (mid[-1] == /*{*/'}') { sp += mid - beg - 1; end = NULL; } else { end = mid + (wdscan(mid, ADELIM) - mid); if (end >= stg) goto unwind_substsyn; end[-2] = EOS; sp += end - beg - 1; } evaluate(substitute(stg = wdstrip(beg, false, false), 0), &from, KSH_UNWIND_ERROR, true); afree(stg, ATEMP); if (end) { evaluate(substitute(stg = wdstrip(mid, false, false), 0), &num, KSH_UNWIND_ERROR, true); afree(stg, ATEMP); } afree(beg, ATEMP); beg = str_val(st->var); flen = utflen(beg); if (from < 0) { if (-from < flen) finc = flen + from; } else finc = from < flen ? from : flen; if (UTFMODE) utfincptr(beg, &finc); beg += finc; flen = utflen(beg); if (num < 0 || num > flen) num = flen; if (UTFMODE) utfincptr(beg, &num); strndupx(x.str, beg, num, ATEMP); goto do_CSUBST; } case '/': { char *s, *p, *d, *sbeg, *end; char *pat, *rrep; char *tpat0, *tpat1, *tpat2; s = wdcopy(sp, ATEMP); p = s + (wdscan(sp, ADELIM) - sp); d = s + (wdscan(sp, CSUBST) - sp); if (p >= d) goto unwind_substsyn; p[-2] = EOS; if (p[-1] == /*{*/'}') d = NULL; else d[-2] = EOS; sp += (d ? d : p) - s - 1; tpat0 = wdstrip(s, true, true); pat = substitute(tpat0, 0); if (d) { d = wdstrip(p, true, false); rrep = substitute(d, 0); afree(d, ATEMP); } else rrep = null; afree(s, ATEMP); s = d = pat; while (*s) if (*s != '\\' || s[1] == '%' || s[1] == '#' || s[1] == '\0' || /* XXX really? */ s[1] == '\\' || s[1] == '/') *d++ = *s++; else s++; *d = '\0'; afree(tpat0, ATEMP); /* reject empty pattern */ if (!*pat || gmatchx("", pat, false)) goto no_repl; /* prepare string on which to work */ strdupx(s, str_val(st->var), ATEMP); sbeg = s; /* first see if we have any match at all */ tpat0 = pat; if (*pat == '#') { /* anchor at the beginning */ tpat1 = shf_smprintf("%s%c*", ++tpat0, MAGIC); tpat2 = tpat1; } else if (*pat == '%') { /* anchor at the end */ tpat1 = shf_smprintf("%c*%s", MAGIC, ++tpat0); tpat2 = tpat0; } else { /* float */ tpat1 = shf_smprintf("%c*%s%c*", MAGIC, pat, MAGIC); tpat2 = tpat1 + 2; } again_repl: /* this would not be necessary if gmatchx would return * the start and end values of a match found, like re* */ if (!gmatchx(sbeg, tpat1, false)) goto end_repl; end = strnul(s); /* now anchor the beginning of the match */ if (*pat != '#') while (sbeg <= end) { if (gmatchx(sbeg, tpat2, false)) break; else sbeg++; } /* now anchor the end of the match */ p = end; if (*pat != '%') while (p >= sbeg) { bool gotmatch; c = *p; *p = '\0'; gotmatch = gmatchx(sbeg, tpat0, false); *p = c; if (gotmatch) break; p--; } strndupx(end, s, sbeg - s, ATEMP); d = shf_smprintf("%s%s%s", end, rrep, p); afree(end, ATEMP); sbeg = d + (sbeg - s) + strlen(rrep); afree(s, ATEMP); s = d; if (stype & 0x80) goto again_repl; end_repl: afree(tpat1, ATEMP); x.str = s; no_repl: afree(pat, ATEMP); if (rrep != null) afree(rrep, ATEMP); goto do_CSUBST; } case '#': case '%': /* ! DOBLANK,DOBRACE_,DOTILDE */ f = DOPAT | (f&DONTRUNCOMMAND) | DOTEMP_; quote = 0; /* Prepend open pattern (so | * in a trim will work as * expected) */ *dp++ = MAGIC; *dp++ = (char)('@' | 0x80); break; case '=': /* Enabling tilde expansion * after :s here is * non-standard ksh, but is * consistent with rules for * other assignments. Not * sure what POSIX thinks of * this. * Not doing tilde expansion * for integer variables is a * non-POSIX thing - makes * sense though, since ~ is * a arithmetic operator. */ if (!(x.var->flag & INTEGER)) f |= DOASNTILDE|DOTILDE; f |= DOTEMP_; /* These will be done after the * value has been assigned. */ f &= ~(DOBLANK|DOGLOB|DOBRACE_); tilde_ok = 1; break; case '?': f &= ~DOBLANK; f |= DOTEMP_; /* FALLTHROUGH */ default: /* Enable tilde expansion */ tilde_ok = 1; f |= DOTILDE; } } else /* skip word */ sp += wdscan(sp, CSUBST) - sp; continue; } case CSUBST: /* only get here if expanding word */ do_CSUBST: sp++; /* ({) skip the } or x */ tilde_ok = 0; /* in case of ${unset:-} */ *dp = '\0'; quote = st->quote; f = st->f; if (f&DOBLANK) doblank--; switch (st->stype&0x7f) { case '#': case '%': /* Append end-pattern */ *dp++ = MAGIC; *dp++ = ')'; *dp = '\0'; dp = Xrestpos(ds, dp, st->base); /* Must use st->var since calling * global would break things * like x[i+=1]. */ x.str = trimsub(str_val(st->var), dp, st->stype); type = strlen(x.str) ? XSUB : XNULLSUB; if (f&DOBLANK) doblank++; st = st->prev; continue; case '=': /* Restore our position and substitute * the value of st->var (may not be * the assigned value in the presence * of integer/right-adj/etc attributes). */ dp = Xrestpos(ds, dp, st->base); /* Must use st->var since calling * global would cause with things * like x[i+=1] to be evaluated twice. */ /* Note: not exported by FEXPORT * in AT&T ksh. */ /* XXX POSIX says readonly is only * fatal for special builtins (setstr * does readonly check). */ len = strlen(dp) + 1; setstr(st->var, debunk(alloc(len, ATEMP), dp, len), KSH_UNWIND_ERROR); x.str = str_val(st->var); type = XSUB; if (f&DOBLANK) doblank++; st = st->prev; continue; case '?': { char *s = Xrestpos(ds, dp, st->base); errorf("%s: %s", st->var->name, dp == s ? "parameter null or not set" : (debunk(s, s, strlen(s) + 1), s)); } case '0': case '/': dp = Xrestpos(ds, dp, st->base); type = XSUB; if (f&DOBLANK) doblank++; st = st->prev; continue; } st = st->prev; type = XBASE; continue; case OPAT: /* open pattern: *(foo|bar) */ /* Next char is the type of pattern */ make_magic = 1; c = *sp++ + 0x80; break; case SPAT: /* pattern separator (|) */ make_magic = 1; c = '|'; break; case CPAT: /* close pattern */ make_magic = 1; c = /*(*/ ')'; break; } break; case XNULLSUB: /* Special case for "$@" (and "${foo[@]}") - no * word is generated if $# is 0 (unless there is * other stuff inside the quotes). */ type = XBASE; if (f&DOBLANK) { doblank--; /* not really correct: x=; "$x$@" should * generate a null argument and * set A; "${@:+}" shouldn't. */ if (dp == Xstring(ds, dp)) word = IFS_WS; } continue; case XSUB: case XSUBMID: if ((c = *x.str++) == 0) { type = XBASE; if (f&DOBLANK) doblank--; continue; } break; case XARGSEP: type = XARG; quote = 1; case XARG: if ((c = *x.str++) == '\0') { /* force null words to be created so * set -- '' 2 ''; foo "$@" will do * the right thing */ if (quote && x.split) word = IFS_WORD; if ((x.str = *x.u.strv++) == NULL) { type = XBASE; if (f&DOBLANK) doblank--; continue; } c = ifs0; if (c == 0) { if (quote && !x.split) continue; c = ' '; } if (quote && x.split) { /* terminate word for "$@" */ type = XARGSEP; quote = 0; } } break; case XCOM: if (newlines) { /* Spit out saved NLs */ c = '\n'; --newlines; } else { while ((c = shf_getc(x.u.shf)) == 0 || c == '\n') if (c == '\n') /* Save newlines */ newlines++; if (newlines && c != EOF) { shf_ungetc(c, x.u.shf); c = '\n'; --newlines; } } if (c == EOF) { newlines = 0; shf_close(x.u.shf); if (x.split) subst_exstat = waitlast(); type = XBASE; if (f&DOBLANK) doblank--; continue; } break; } /* check for end of word or IFS separation */ if (c == 0 || (!quote && (f & DOBLANK) && doblank && !make_magic && ctype(c, C_IFS))) { /* How words are broken up: * | value of c * word | ws nws 0 * ----------------------------------- * IFS_WORD w/WS w/NWS w * IFS_WS -/WS w/NWS - * IFS_NWS -/NWS w/NWS w * (w means generate a word) * Note that IFS_NWS/0 generates a word (AT&T ksh * doesn't do this, but POSIX does). */ if (word == IFS_WORD || (!ctype(c, C_IFSWS) && c && word == IFS_NWS)) { char *p; *dp++ = '\0'; p = Xclose(ds, dp); if (fdo & DOBRACE_) /* also does globbing */ alt_expand(wp, p, p, p + Xlength(ds, (dp - 1)), fdo | (f & DOMARKDIRS)); else if (fdo & DOGLOB) glob(p, wp, f & DOMARKDIRS); else if ((f & DOPAT) || !(fdo & DOMAGIC_)) XPput(*wp, p); else XPput(*wp, debunk(p, p, strlen(p) + 1)); fdo = 0; saw_eq = 0; tilde_ok = (f & (DOTILDE|DOASNTILDE)) ? 1 : 0; if (c != 0) Xinit(ds, dp, 128, ATEMP); } if (c == 0) return; if (word != IFS_NWS) word = ctype(c, C_IFSWS) ? IFS_WS : IFS_NWS; } else { if (type == XSUB) { if (word == IFS_NWS && Xlength(ds, dp) == 0) { char *p; *(p = alloc(1, ATEMP)) = '\0'; XPput(*wp, p); } type = XSUBMID; } /* age tilde_ok info - ~ code tests second bit */ tilde_ok <<= 1; /* mark any special second pass chars */ if (!quote) switch (c) { case '[': case NOT: case '-': case ']': /* For character classes - doesn't hurt * to have magic !,-,]s outside of * [...] expressions. */ if (f & (DOPAT | DOGLOB)) { fdo |= DOMAGIC_; if (c == '[') fdo |= f & DOGLOB; *dp++ = MAGIC; } break; case '*': case '?': if (f & (DOPAT | DOGLOB)) { fdo |= DOMAGIC_ | (f & DOGLOB); *dp++ = MAGIC; } break; case OBRACE: case ',': case CBRACE: if ((f & DOBRACE_) && (c == OBRACE || (fdo & DOBRACE_))) { fdo |= DOBRACE_|DOMAGIC_; *dp++ = MAGIC; } break; case '=': /* Note first unquoted = for ~ */ if (!(f & DOTEMP_) && !saw_eq && (Flag(FBRACEEXPAND) || (f & DOASNTILDE))) { saw_eq = 1; tilde_ok = 1; } break; case ':': /* : */ /* Note unquoted : for ~ */ if (!(f & DOTEMP_) && (f & DOASNTILDE)) tilde_ok = 1; break; case '~': /* tilde_ok is reset whenever * any of ' " $( $(( ${ } are seen. * Note that tilde_ok must be preserved * through the sequence ${A=a=}~ */ if (type == XBASE && (f & (DOTILDE|DOASNTILDE)) && (tilde_ok & 2)) { const char *p; char *dp_x; dp_x = dp; p = maybe_expand_tilde(sp, &ds, &dp_x, f & DOASNTILDE); if (p) { if (dp != dp_x) word = IFS_WORD; dp = dp_x; sp = p; continue; } } break; } else quote &= ~2; /* undo temporary */ if (make_magic) { make_magic = 0; fdo |= DOMAGIC_ | (f & DOGLOB); *dp++ = MAGIC; } else if (ISMAGIC(c)) { fdo |= DOMAGIC_; *dp++ = MAGIC; } *dp++ = c; /* save output char */ word = IFS_WORD; } } } /* * Prepare to generate the string returned by ${} substitution. */ static int varsub(Expand *xp, const char *sp, const char *word, int *stypep, /* becomes qualifier type */ int *slenp) /* " " len (=, :=, etc.) valid iff *stypep != 0 */ { int c; int state; /* next state: XBASE, XARG, XSUB, XNULLSUB */ int stype; /* substitution type */ int slen; const char *p; struct tbl *vp; bool zero_ok = false; if ((stype = sp[0]) == '\0') /* Bad variable name */ return (-1); xp->var = NULL; /*- * ${#var}, string length (-U: characters, +U: octets) or array size * ${%var}, string width (-U: screen columns, +U: octets) */ c = sp[1]; if (stype == '%' && c == '\0') return (-1); if ((stype == '#' || stype == '%') && c != '\0') { /* Can't have any modifiers for ${#...} or ${%...} */ if (*word != CSUBST) return (-1); sp++; /* Check for size of array */ if ((p = cstrchr(sp, '[')) && (p[1] == '*' || p[1] == '@') && p[2] == ']') { int n = 0; if (stype != '#') return (-1); vp = global(arrayname(sp)); if (vp->flag & (ISSET|ARRAY)) zero_ok = true; for (; vp; vp = vp->u.array) if (vp->flag & ISSET) n++; c = n; } else if (c == '*' || c == '@') { if (stype != '#') return (-1); c = e->loc->argc; } else { p = str_val(global(sp)); zero_ok = p != null; if (stype == '#') c = utflen(p); else { /* partial utf_mbswidth reimplementation */ const char *s = p; unsigned int wc; size_t len; int cw; c = 0; while (*s) { if (!UTFMODE || (len = utf_mbtowc(&wc, s)) == (size_t)-1) /* not UTFMODE or not UTF-8 */ wc = (unsigned char)(*s++); else /* UTFMODE and UTF-8 */ s += len; /* wc == char or wchar at s++ */ if ((cw = utf_wcwidth(wc)) == -1) { /* 646, 8859-1, 10646 C0/C1 */ c = -1; break; } c += cw; } } } if (Flag(FNOUNSET) && c == 0 && !zero_ok) errorf("%s: parameter not set", sp); *stypep = 0; /* unqualified variable/string substitution */ xp->str = shf_smprintf("%d", c); return (XSUB); } /* Check for qualifiers in word part */ stype = 0; c = word[slen = 0] == CHAR ? word[1] : 0; if (c == ':') { slen += 2; stype = 0x80; c = word[slen + 0] == CHAR ? word[slen + 1] : 0; } if (!stype && c == '/') { slen += 2; stype = c; if (word[slen] == ADELIM) { slen += 2; stype |= 0x80; } } else if (stype == 0x80 && (c == ' ' || c == '0')) { stype |= '0'; } else if (ctype(c, C_SUBOP1)) { slen += 2; stype |= c; } else if (ctype(c, C_SUBOP2)) { /* Note: ksh88 allows :%, :%%, etc */ slen += 2; stype = c; if (word[slen + 0] == CHAR && c == word[slen + 1]) { stype |= 0x80; slen += 2; } } else if (stype) /* : is not ok */ return (-1); if (!stype && *word != CSUBST) return (-1); *stypep = stype; *slenp = slen; c = sp[0]; if (c == '*' || c == '@') { switch (stype & 0x7f) { case '=': /* can't assign to a vector */ case '%': /* can't trim a vector (yet) */ case '#': return (-1); } if (e->loc->argc == 0) { xp->str = null; xp->var = global(sp); state = c == '@' ? XNULLSUB : XSUB; } else { xp->u.strv = (const char **)e->loc->argv + 1; xp->str = *xp->u.strv++; xp->split = c == '@'; /* $@ */ state = XARG; } zero_ok = true; /* POSIX 2009? */ } else { if ((p = cstrchr(sp, '[')) && (p[1] == '*' || p[1] == '@') && p[2] == ']') { XPtrV wv; switch (stype & 0x7f) { case '=': /* can't assign to a vector */ case '%': /* can't trim a vector (yet) */ case '#': case '?': return (-1); } XPinit(wv, 32); if ((c = sp[0]) == '!') ++sp; vp = global(arrayname(sp)); for (; vp; vp = vp->u.array) { if (!(vp->flag&ISSET)) continue; XPput(wv, c == '!' ? shf_smprintf("%lu", arrayindex(vp)) : str_val(vp)); } if (XPsize(wv) == 0) { xp->str = null; state = p[1] == '@' ? XNULLSUB : XSUB; XPfree(wv); } else { XPput(wv, 0); xp->u.strv = (const char **)XPptrv(wv); xp->str = *xp->u.strv++; xp->split = p[1] == '@'; /* ${foo[@]} */ state = XARG; } } else { /* Can't assign things like $! or $1 */ if ((stype & 0x7f) == '=' && ctype(*sp, C_VAR1 | C_DIGIT)) return (-1); if (*sp == '!' && sp[1]) { ++sp; xp->var = global(sp); if (cstrchr(sp, '[')) { if (xp->var->flag & ISSET) xp->str = shf_smprintf("%lu", arrayindex(xp->var)); else xp->str = null; } else if (xp->var->flag & ISSET) xp->str = xp->var->name; else xp->str = "0"; /* ksh93 compat */ } else { xp->var = global(sp); xp->str = str_val(xp->var); } state = XSUB; } } c = stype&0x7f; /* test the compiler's code generator */ if (ctype(c, C_SUBOP2) || stype == (0x80 | '0') || c == '/' || (((stype&0x80) ? *xp->str=='\0' : xp->str==null) ? /* undef? */ c == '=' || c == '-' || c == '?' : c == '+')) state = XBASE; /* expand word instead of variable value */ if (Flag(FNOUNSET) && xp->str == null && !zero_ok && (ctype(c, C_SUBOP2) || (state != XBASE && c != '+'))) errorf("%s: parameter not set", sp); return (state); } /* * Run the command in $(...) and read its output. */ static int comsub(Expand *xp, const char *cp) { Source *s, *sold; struct op *t; struct shf *shf; s = pushs(SSTRING, ATEMP); s->start = s->str = cp; sold = source; t = compile(s); afree(s, ATEMP); source = sold; if (t == NULL) return (XBASE); if (t != NULL && t->type == TCOM && /* $(args == NULL && *t->vars == NULL && t->ioact != NULL) { struct ioword *io = *t->ioact; char *name; if ((io->flag&IOTYPE) != IOREAD) errorf("funny $() command: %s", snptreef(NULL, 32, "%R", io)); shf = shf_open(name = evalstr(io->name, DOTILDE), O_RDONLY, 0, SHF_MAPHI|SHF_CLEXEC); if (shf == NULL) errorf("%s: cannot open $() input", name); xp->split = 0; /* no waitlast() */ } else { int ofd1, pv[2]; openpipe(pv); shf = shf_fdopen(pv[0], SHF_RD, NULL); ofd1 = savefd(1); if (pv[1] != 1) { ksh_dup2(pv[1], 1, false); close(pv[1]); } execute(t, XFORK|XXCOM|XPIPEO, NULL); restfd(1, ofd1); startlast(); xp->split = 1; /* waitlast() */ } xp->u.shf = shf; return (XCOM); } /* * perform #pattern and %pattern substitution in ${} */ static char * trimsub(char *str, char *pat, int how) { char *end = strnul(str); char *p, c; switch (how & 0xFF) { case '#': /* shortest at beginning */ for (p = str; p <= end; p += utf_ptradj(p)) { c = *p; *p = '\0'; if (gmatchx(str, pat, false)) { *p = c; return (p); } *p = c; } break; case '#'|0x80: /* longest match at beginning */ for (p = end; p >= str; p--) { c = *p; *p = '\0'; if (gmatchx(str, pat, false)) { *p = c; return (p); } *p = c; } break; case '%': /* shortest match at end */ p = end; while (p >= str) { if (gmatchx(p, pat, false)) goto trimsub_match; if (UTFMODE) { char *op = p; while ((p-- > str) && ((*p & 0xC0) == 0x80)) ; if ((p < str) || (p + utf_ptradj(p) != op)) p = op - 1; } else --p; } break; case '%'|0x80: /* longest match at end */ for (p = str; p <= end; p++) if (gmatchx(p, pat, false)) { trimsub_match: strndupx(end, str, p - str, ATEMP); return (end); } break; } return (str); /* no match, return string */ } /* * glob * Name derived from V6's /etc/glob, the program that expanded filenames. */ /* XXX cp not const 'cause slashes are temporarily replaced with NULs... */ static void glob(char *cp, XPtrV *wp, int markdirs) { int oldsize = XPsize(*wp); if (glob_str(cp, wp, markdirs) == 0) XPput(*wp, debunk(cp, cp, strlen(cp) + 1)); else qsort(XPptrv(*wp) + oldsize, XPsize(*wp) - oldsize, sizeof(void *), xstrcmp); } #define GF_NONE 0 #define GF_EXCHECK BIT(0) /* do existence check on file */ #define GF_GLOBBED BIT(1) /* some globbing has been done */ #define GF_MARKDIR BIT(2) /* add trailing / to directories */ /* Apply file globbing to cp and store the matching files in wp. Returns * the number of matches found. */ int glob_str(char *cp, XPtrV *wp, int markdirs) { int oldsize = XPsize(*wp); XString xs; char *xp; Xinit(xs, xp, 256, ATEMP); globit(&xs, &xp, cp, wp, markdirs ? GF_MARKDIR : GF_NONE); Xfree(xs, xp); return (XPsize(*wp) - oldsize); } static void globit(XString *xs, /* dest string */ char **xpp, /* ptr to dest end */ char *sp, /* source path */ XPtrV *wp, /* output list */ int check) /* GF_* flags */ { char *np; /* next source component */ char *xp = *xpp; char *se; char odirsep; /* This to allow long expansions to be interrupted */ intrcheck(); if (sp == NULL) { /* end of source path */ /* We only need to check if the file exists if a pattern * is followed by a non-pattern (eg, foo*x/bar; no check * is needed for foo* since the match must exist) or if * any patterns were expanded and the markdirs option is set. * Symlinks make things a bit tricky... */ if ((check & GF_EXCHECK) || ((check & GF_MARKDIR) && (check & GF_GLOBBED))) { #define stat_check() (stat_done ? stat_done : \ (stat_done = stat(Xstring(*xs, xp), &statb) < 0 \ ? -1 : 1)) struct stat lstatb, statb; int stat_done = 0; /* -1: failed, 1 ok */ if (lstat(Xstring(*xs, xp), &lstatb) < 0) return; /* special case for systems which strip trailing * slashes from regular files (eg, /etc/passwd/). * SunOS 4.1.3 does this... */ if ((check & GF_EXCHECK) && xp > Xstring(*xs, xp) && xp[-1] == '/' && !S_ISDIR(lstatb.st_mode) && (!S_ISLNK(lstatb.st_mode) || stat_check() < 0 || !S_ISDIR(statb.st_mode))) return; /* Possibly tack on a trailing / if there isn't already * one and if the file is a directory or a symlink to a * directory */ if (((check & GF_MARKDIR) && (check & GF_GLOBBED)) && xp > Xstring(*xs, xp) && xp[-1] != '/' && (S_ISDIR(lstatb.st_mode) || (S_ISLNK(lstatb.st_mode) && stat_check() > 0 && S_ISDIR(statb.st_mode)))) { *xp++ = '/'; *xp = '\0'; } } strndupx(np, Xstring(*xs, xp), Xlength(*xs, xp), ATEMP); XPput(*wp, np); return; } if (xp > Xstring(*xs, xp)) *xp++ = '/'; while (*sp == '/') { Xcheck(*xs, xp); *xp++ = *sp++; } np = strchr(sp, '/'); if (np != NULL) { se = np; odirsep = *np; /* don't assume '/', can be multiple kinds */ *np++ = '\0'; } else { odirsep = '\0'; /* keep gcc quiet */ se = sp + strlen(sp); } /* Check if sp needs globbing - done to avoid pattern checks for strings * containing MAGIC characters, open [s without the matching close ], * etc. (otherwise opendir() will be called which may fail because the * directory isn't readable - if no globbing is needed, only execute * permission should be required (as per POSIX)). */ if (!has_globbing(sp, se)) { XcheckN(*xs, xp, se - sp + 1); debunk(xp, sp, Xnleft(*xs, xp)); xp += strlen(xp); *xpp = xp; globit(xs, xpp, np, wp, check); } else { DIR *dirp; struct dirent *d; char *name; int len; int prefix_len; /* xp = *xpp; copy_non_glob() may have re-alloc'd xs */ *xp = '\0'; prefix_len = Xlength(*xs, xp); dirp = opendir(prefix_len ? Xstring(*xs, xp) : "."); if (dirp == NULL) goto Nodir; while ((d = readdir(dirp)) != NULL) { name = d->d_name; if (name[0] == '.' && (name[1] == 0 || (name[1] == '.' && name[2] == 0))) continue; /* always ignore . and .. */ if ((*name == '.' && *sp != '.') || !gmatchx(name, sp, true)) continue; len = strlen(d->d_name) + 1; XcheckN(*xs, xp, len); memcpy(xp, name, len); *xpp = xp + len - 1; globit(xs, xpp, np, wp, (check & GF_MARKDIR) | GF_GLOBBED | (np ? GF_EXCHECK : GF_NONE)); xp = Xstring(*xs, xp) + prefix_len; } closedir(dirp); Nodir: ; } if (np != NULL) *--np = odirsep; } /* remove MAGIC from string */ char * debunk(char *dp, const char *sp, size_t dlen) { char *d; const char *s; if ((s = cstrchr(sp, MAGIC))) { if (s - sp >= (ssize_t)dlen) return (dp); memmove(dp, sp, s - sp); for (d = dp + (s - sp); *s && (d - dp < (ssize_t)dlen); s++) if (!ISMAGIC(*s) || !(*++s & 0x80) || !vstrchr("*+?@! ", *s & 0x7f)) *d++ = *s; else { /* extended pattern operators: *+?@! */ if ((*s & 0x7f) != ' ') *d++ = *s & 0x7f; if (d - dp < (ssize_t)dlen) *d++ = '('; } *d = '\0'; } else if (dp != sp) strlcpy(dp, sp, dlen); return (dp); } /* Check if p is an unquoted name, possibly followed by a / or :. If so * puts the expanded version in *dcp,dp and returns a pointer in p just * past the name, otherwise returns 0. */ static const char * maybe_expand_tilde(const char *p, XString *dsp, char **dpp, int isassign) { XString ts; char *dp = *dpp; char *tp; const char *r; Xinit(ts, tp, 16, ATEMP); /* : only for DOASNTILDE form */ while (p[0] == CHAR && p[1] != '/' && (!isassign || p[1] != ':')) { Xcheck(ts, tp); *tp++ = p[1]; p += 2; } *tp = '\0'; r = (p[0] == EOS || p[0] == CHAR || p[0] == CSUBST) ? tilde(Xstring(ts, tp)) : NULL; Xfree(ts, tp); if (r) { while (*r) { Xcheck(*dsp, dp); if (ISMAGIC(*r)) *dp++ = MAGIC; *dp++ = *r++; } *dpp = dp; r = p; } return (r); } /* * tilde expansion * * based on a version by Arnold Robbins */ static char * tilde(char *cp) { char *dp = null; if (cp[0] == '\0') dp = str_val(global("HOME")); else if (cp[0] == '+' && cp[1] == '\0') dp = str_val(global("PWD")); else if (cp[0] == '-' && cp[1] == '\0') dp = str_val(global("OLDPWD")); #ifndef MKSH_NOPWNAM else dp = homedir(cp); #endif /* If HOME, PWD or OLDPWD are not set, don't expand ~ */ return (dp == null ? NULL : dp); } #ifndef MKSH_NOPWNAM /* * map userid to user's home directory. * note that 4.3's getpw adds more than 6K to the shell, * and the YP version probably adds much more. * we might consider our own version of getpwnam() to keep the size down. */ static char * homedir(char *name) { struct tbl *ap; ap = ktenter(&homedirs, name, hash(name)); if (!(ap->flag & ISSET)) { struct passwd *pw; pw = getpwnam(name); if (pw == NULL) return (NULL); strdupx(ap->val.s, pw->pw_dir, APERM); ap->flag |= DEFINED|ISSET|ALLOC; } return (ap->val.s); } #endif static void alt_expand(XPtrV *wp, char *start, char *exp_start, char *end, int fdo) { int count = 0; char *brace_start, *brace_end, *comma = NULL; char *field_start; char *p; /* search for open brace */ for (p = exp_start; (p = strchr(p, MAGIC)) && p[1] != OBRACE; p += 2) ; brace_start = p; /* find matching close brace, if any */ if (p) { comma = NULL; count = 1; for (p += 2; *p && count; p++) { if (ISMAGIC(*p)) { if (*++p == OBRACE) count++; else if (*p == CBRACE) --count; else if (*p == ',' && count == 1) comma = p; } } } /* no valid expansions... */ if (!p || count != 0) { /* Note that given a{{b,c} we do not expand anything (this is * what AT&T ksh does. This may be changed to do the {b,c} * expansion. } */ if (fdo & DOGLOB) glob(start, wp, fdo & DOMARKDIRS); else XPput(*wp, debunk(start, start, end - start)); return; } brace_end = p; if (!comma) { alt_expand(wp, start, brace_end, end, fdo); return; } /* expand expression */ field_start = brace_start + 2; count = 1; for (p = brace_start + 2; p != brace_end; p++) { if (ISMAGIC(*p)) { if (*++p == OBRACE) count++; else if ((*p == CBRACE && --count == 0) || (*p == ',' && count == 1)) { char *new; int l1, l2, l3; l1 = brace_start - start; l2 = (p - 1) - field_start; l3 = end - brace_end; new = alloc(l1 + l2 + l3 + 1, ATEMP); memcpy(new, start, l1); memcpy(new + l1, field_start, l2); memcpy(new + l1 + l2, brace_end, l3); new[l1 + l2 + l3] = '\0'; alt_expand(wp, new, new + l1, new + l1 + l2 + l3, fdo); field_start = p + 1; } } } return; }